Stealth Announces PULSAR-VLDS in collaboration with Virginia Longitudinal Data System (VLDS)
Los Angeles, CA – Stealth Software Technologies, Inc., in collaboration with VLDS (Virginia Longitudinal Data System), a pioneering collaboration among Virginia state agencies to produce better data for public policy, announces the release of a White Paper on “PULSAR-VLDS”, an innovative privacy-enhancing technology system that enables data sharings and analytics across multiple datasets with strong data privacy guarantees. PULSAR-VLDS is a federated technology system designed to extract, shape, and analyze data from multiple Virginia government partner agencies without requiring the agencies to disclose de-identified individual data to other agencies.
Why should you care? A vast amount of data is generated and accumulated with more and more required for comprehensive research, but this data is not being used to its full potential due to legal and regulatory restrictions. For example, evidence-based policy-makers require cross-category reports that often require accessing, extracting, and analyzing datasets held by distinct government branches and agencies. However, the data disclosure postures dictated by these agencies usually prohibit such comprehensive data analytic research, damaging the quality of decision makings. Our technology aims to achieve the outcomes of the report while simultaneously protecting agency data.
How it works. PULSAR-VLDS leverages its novel multiparty computation (MPC) engine that enables data owning agencies to generate data analytics reports by allowing secure data join and aggregate computation across multiple datasets without ever disclosing the plaintext data (not even noisy or de-identified) to others. Our PULSAR-VLDS engine is installed on-prem for participating data owner agencies, and seamlessly integrates with existing database systems. It operates on streams of datasets, whereupon it encrypts and processes data on the fly, hence never requiring a large memory footprint. The output data analytics report contains only the requested, limited aggregates and individual data is never included.
Find out more. We conducted a research experiment with VLDS using our software. Our experimentation was conducted with VLDS standard datasets (Wage and School datasets) and a client query, which requests an analytics report with an average and standard deviation on wages in 2018 who graduated with a bachelor’s degree by all genders and institutions. In the experimentations, our system securely joins and computes the requested statistics across two datasets of distinct sizes in total, ranging from 40K records to 1.58M records, demonstrating the 0.85 milliseconds for the end-to-end latency per record. Read about the project and the PULSAR-VLDS White Paper at https://www.stealthsoftwareinc.com/solutions/projects/pulsar/vlds/
“The PULSAR-VLDS is a significant step in creating a cutting-edge cryptography tool, known as secure multiparty computation (MPC) that addresses the tension between protecting personal data and maximizing the ability to share data. The PULSAR-VLDS MPC engine demonstrates the feasibility of a new privacy enhancing technology enabling state policy analysis at the core of the VLDS mission. Agencies and citizens can have much greater assurance in the security and confidentiality of citizen data while meeting the essential needs of policymakers.”
– Tod R. Massa, Director, Policy Analytics, VLDS (Virginia Longitudinal Data System) Lead